The manager-of-one weekly review.

You manage people, projects, agents — and yourself. That last report is the one nobody schedules a 1:1 with. This is that 1:1. The format is the FIT loop — Feel, Inventory, Target — which is just TRAM pointed at the mirror: Track what happened, Report it honestly to yourself, Align on the one thing that matters next, Measure whether it moved. An org of one, run on the same loop as the org of a thousand.

One rule makes it work: you always start by closing last week's loop. Did last week's target happen — observably? That question, asked every week, is the whole compounding engine. Intentions don't compound. Closed loops do.

Run it the same time every week — Friday afternoon is ours; the week is fresh and the weekend is a natural boundary — on paper or with your AI partner, and don't let it grow past fifteen minutes. A review that takes an hour gets skipped by week three. A review that takes fifteen minutes is still running in year five — and year five is where the compounding lives.

Print it, or paste it into your AI partner. The times in parentheses are the budget — if a step runs long, you're overthinking it.

MANAGER-OF-ONE WEEKLY REVIEW                        Week of: [ ]

0. CLOSE LAST WEEK'S LOOP  (2 min)  — the step everything compounds on
   Last week's target: [ ]
   Did it happen — observably?          [ ] yes   [ ] no
   If no: what actually got in the way? (Name the real thing.)

1. FEEL  (3 min)
   What has my stomach in a knot this week?
   Not the surface ("the project is slipping") — the real thing
   ("I'm avoiding a conversation with the person slipping it").
   The knot: ________________________________________________
   What it's really about: __________________________________

2. INVENTORY  (5 min)
   Ground truth, not the story I prefer:
   - Time — where did the week actually go? Biggest surprise:
   - Wins — up to 3, with receipts (result, not effort):
     1.
     2.
     3.
   - Drains — up to 2 recurring things that cost more than they
     return (a meeting, a task, a habit):
     1.
     2.
   - Constraints that are real this week (energy, politics, skill):

3. TARGET  (5 min)
   ONE observable change for next week. One.
   Small enough to be unmissable. Checkable by someone else.
   Not "communicate better" — "have the conversation by Wednesday."
   Target: __________________________________________________
   By when: ____________  Visible to whom: __________________

Once a month, append two questions. Which one stakeholder should have seen my impact this month — and did they? (Visibility is managed, not wished for.) And which drain do I retire permanently — delegate it, automate it, brief an agent on it, or kill it.

Once a year, the weekly reviews become your retro — and your raise case. Being busy is not the same as being valuable, and no one is keeping track for you; but if you've run this loop, you have. Harvest it: three concrete results with receipts, one skill that levels you up next year, two drains you eliminated, one sentence of direction for the year ahead. Review season stops being a memory test and becomes a reading of the record.

If an agent holds your journal — meetings, commitments, what shipped — the inventory step stops relying on your memory, which is the step where memory lies most. (If you built the Work OS, the agent already has your journal.) Paste this standing brief — which is itself an agent brief:

Every Friday at [time], run my Manager-of-One review:
1. Show last week's target and any evidence it did/didn't happen.
2. Draft my inventory from the record: where my time went, up to
   3 wins WITH receipts, up to 2 recurring drains. Truth over tidy —
   don't make me sound busier or better than the record supports.
3. Ask me the Feel question. Don't answer it for me. Wait.
4. Push back if my target is vague, plural, or unobservable.
5. File the completed review; keep a running list of closed loops.

Note what stays human: the machine assembles the evidence and keeps you honest; the feeling, the naming of the real knot, and the choice of target are yours. That division of labor — assistant executes, you manage — is the whole thing in miniature.

One target a week feels timid. It is arithmetic. Fifty-two closed loops a year beats twelve abandoned resolutions by a distance that compounds — one percent better every day is not a poster, it is a schedule.

The reviews that change careers are not the profound ones. They are the ones that happen.

The part that makes them happen — assembling the honest inventory, week after week — happens to be the part Ari keeps for you.